Design And Study Of Cloud Service Monitoring System And Its Status Early Warning Scheme

The rapid development of Internet of Things technologies provides many conveniences for people's lives.Cloud services are important guarantee to support various IoT applications.However,with the massive access of IoT devices and the rapid increase in the number of users,the difficulty and cost of cloud platform maintenance continues to rise.The diversification and complexity of cloud services has become an inevitable trend of development.Therefore,in order to guarantee the quality of services and improve the stability of services,it is necessary to implement more efficient and complete monitoring of cloud services to ensure that managers can grasp the abnormal situations and illegal attacks in real time.This topic aims to provide insight about the design and implement of the monitoring system for cloud services.In addition,researching on the efficient early warning strategy is another focus of the thesis.Firstly,according to the characteristics of cloud services and the needs of security early warning,the overall architecture of cloud service monitoring system is designed and modularized into five components,i.e.,data acquisition module,data forwarding module,data storage module,early warning module,and display module.In view of the existing problems in key technologies of open source monitoring systems,this thesis proposes corresponding solutions.In the data acquisition module,this thesis focuses on the realization of monitoring services discovery and invocation with high scalability.Moreover,a dynamic push acquisition model is proposed for data acquisition.In data forwarding module,the collected monitoring data is verified and transmitted.Kafka is used as a messaging system among modules to guarantee the high performance and reliability.Besides,an improved load balancing algorithm is designed for Kafka.In the data storage module,this thesis proposes a data storage scheme based on MongoDB and implements the"database dual activity"scheme,which ensures the high availability of the module.Secondly,the early warning scheme based on dynamic threshold and fuzzy logic is proposed and applied to the early warning module of the monitoring system.The strategy uses the improved exponential smoothing algorithm to model and analyze the collected monitoring data,and then predicts the possible values of monitoring indicators at the next moment as well as calculating the dynamic threshold interval.Then,with the improved fuzzy logic algorithms,the abnormal degree of service is calculated by the aggregation of abnormal degrees of different monitoring indicators.Finally,the alarm level is evaluated by comparing the abnormal degree with the alarm intervals.Besides,this thesis evaluates and analyses the performance of the strategy for specific cloud service.The results show that using the proposed early warning strategy can obtain more accurate threshold range of monitoring indicators,and can significantly reduce the probability of false alarm and missed alarm,thereby effectively improving the reliability of the alarm.